Structured Light meets DAVID Laserscanner

[MagWeb]

Hi Greg,
what´s about slides as they are in that toy photocams.: You click once and the "slides" on a rotating disk move a certain degree.
Similar to my color filters disk but inside the optics as vary gobo projectors... stepper driven .... should work

Gunter

[WalterMo]

A similar idea like Greg I already had some time ago to create the phase shift:
viewtopic.php?f=3&t=280&start=45

And now we have the possibility to control such a thing by a microcontroller from DAVID via USB / RS232 adapter.

[Simon]

Hi,
I have done some small improvements based on Joes test images. Please download again:

Edit (29th September 2009): The structured light feature is now part of the official release. So please download the latest version here: http://www.david-laserscanner.com/?section=Downloads

Now the simulated line is more realistic and DAVID is able to filter out many outliers.
You will also find a very first pattern projection dialog when you choose "Structured Light Grabber".

Please note: These features are still far from beeing finished. Currently thay are only designed for a projector with a resolution not much lower than 1024x768, 12 pattern projections, and a camera with a resolution around 1024x768. All other resolutions and number of patterns are not tested and will not work correctly.

Please do not hesitate to post your camera shots! It will help us to carry on improving the structured light feature of DAVID.

Is it possible to have one fringe pattern slide projected but accurately rotate the entire slide projector between frames? What kind of step accuracy would be needed or does it have to be linear movement to create the phase shift?

Yes, a rotation of the projector around the projectors focal line should work. But Gunter is right, I think a linear movement of the slide is a better choice.

[MagWeb]

Hi,
@Simon
Thanks for the improved version I´m gonnado some tests today - with a better projector and my Exilim..

@ moving a projector linear to get the phaseshift:

I think moving the projector to get the phase shift works only on a 2D surface. In 3D ithink it may cause wrong data for a projector "simultates" a "rotating movement" - its lines are not parallel. so it should move slower the nearer the object area is to the projector... A linear movement shifts the projection in a 3d parallel way....

- difficult to explain hope you got what i mean.
Gunter

[MagWeb]

Hi Simon,

cann´t get the projections on the second monitor. the dialog sees the second one but the patterns always cover the first screen.
Any need to set a special position of the second monitor or has it to be exactly at 1024x768 ? (I tried also a real monitor instead of my microprojector) - but that may be a special problem of my bootcamp system as well.

Gunter

[joecnc2006]

Hi Simon,

cann´t get the projections on the second monitor. the dialog sees the second one but the patterns always cover the first screen.
Any need to set a special position of the second monitor or has it to be exactly at 1024x768 ? (I tried also a real monitor instead of my microprojector) - but that may be a special problem of my bootcamp system as well.

Gunter

Same here i have the projector as second monitor and the patterns only show on the 1st monitor, and i know my system is setup correctly because of the Identify monitors the Large No. 2 shows on the projector.

[MagWeb]

Hi,

here a result using the without background feature:

first : I did images of an empty corner
>>> reference scan >>>> rotating motion detected
second: I removed the planes and shot an angel

-my M3 microprojector at native VGA (supports interpolated 1024X768)
-unmodified patterns as posted by Simon (implemented pattern dialog could not be used >> see above)
-Logitech Pro 9000 at 960x720

realy much depends on the settings. I had to set contrast rather high to get a line without nervous jumps>>> so the fringe shots appear as BW stripes without grey between....

Gunter

Edit: Allow data behind the planes to get the wings!

[Bongobat]

Hi,

@ Walter
Sorry I was thinking someone already had the idea but I was too lazy too look it up. Can you explain the piezo actuators? Where are these available?

@ Gunter
The toy rotating disc idea is good but the slide needs to be places correctly and the rotating mechanism has to be very exact I would imagine. I was hoping to use the projector as it is but your explination about parallel vs rotation makes sense. Both rotating disc and linear actuator ideas sound possible. Maybe it can be worked out

@ Simon
How would the software handle analog photographic slides as opposed to digital images? Any problems?

Greg

[hal]

Hello,
unfortunately I could to test David with the projector only one time. Now my Camera seems... in holiday
I don't know why, but it don't work any more I hope to find quickly the problem and a solution.

In the meantime, I've done these 10 .gif animations: 5 with a loop animation and 5 with no loop (animation played only one time). The .gif animated images have inside the 12 patterns created by Simon, in the right sequence.
The animation were done at differents fps speeds (1, 2, 4, 8 and 12 fps), so, you can choose which you prefer and which is better for your purpose.

I hope that these .gif images will be seen by all the users on your pc, under Windows and under Mac o.s. . Please confirm if I'm right.

I've done a test with a projector. At now, I don't have PowerPoint installed or other 2D viewer softwares that allow to show the animation in the second monitor view (the projector).
So, I've used the "desktop background image" as viewer
I've opened the control panel of my Desktop, choose the resolution of the second monitor at 1024 x 768 and choose the "Extend the Desktop on second monitor" option.
After I've only choose the .gif image as background of the desktop... and animation was played at full resolution of 1024x768, without problem (ok, the same animated background appears in the first monitor, and this fact can disturb you during work , but this is a temporary solution, waiting for a better solution to project the patterns sequence only on the second monitor/projector).

To capture the images projected on the object/panels is an other chapter
Now, without my camera, I can't test the best solution the catch the images, but probably the best thing is:
- For fast animation of the sequence: to record a video and after grab the images sliding the timeline frame after frame;
- For slow animation of the sequence: take a camera shoot for every projected image.

Bye, Mattia

[WalterMo]

Hi Greg,
Piezo actuators are mainly built of single piezoeletric disks which are stacked to a column. The applied voltage (up to 100V) and the number of disks determine the possible travel of such a device.
Its advantage: Very precise, very fast and rather strong.
Disadvantages: Not cheap and not easy to use because of the rather high voltage.

Easier and much cheaper is to move the slide by a RC servo. They need 5V power and input pulses with a repetition frequency of 50 Hz and variable width of 1 to 2 ms. The width corresponds to the rotation angle of the output shaft (0 to 90°). If a metric 6mm screw (spindle) would be rotated a linear travel of 0.25mm of a female screw mounted at the slide could be achieved.

Is the atmega8 able to generate such pulses?

[Simon]

Thank you for you feedback so far!

cann´t get the projections on the second monitor. the dialog sees the second one but the patterns always cover the first screen. Any need to set a special position of the second monitor or has it to be exactly at 1024x768 ? (I tried also a real monitor instead of my microprojector) - but that may be a special problem of my bootcamp system as well.

Same here i have the projector as second monitor and the patterns only show on the 1st monitor, and i know my system is setup correctly because of the Identify monitors the Large No. 2 shows on the projector.

Thanks, I found a bug, now it should work. The window should cover the entire screen of the choosen Monitor. Any screen resolution and any position should work. But the user settings are still not saved and restored after restart. Please download again:

DAVID_2_4_1_CLA.zip

@ Simon
How would the software handle analog photographic slides as opposed to digital images? Any problems?

I guess photographic slides should work too. Whether digital or analog, you always need correct gray values. The sine pattern should be projected as correct sine (which is not always the case, since it depends on the quality of the projectors color calibration or on the quality of the photographic slides).

[MagWeb]

Hi Simon,

I debuged my patterngenerator today. The reason for the "nervous" resulting line was a wrong order in the fringes sequence.
To test different resolutions, I made the sin-height configurable - according to the used number of patterns ,too
So the fringes height should be the screen height % the n.th potenz of 2 , where "n" is the number of patterns minus 6 (4 fringes patterns plus 2 enlighted - not enlighted)

So if one uses 12 patterns one needs 64 sin waves (2x2x2x2x2x2)to get the same height of two waves as the stripes of the first Gray pattern (screenheight/32).
So if one uses 11 patterns one needs 32 sin waves (2x2x2x2x2)to get the same height of two waves as the stripes of the second ,which is here the first, Gray pattern (screenheight/16).

Now, running your pattern generator I get 33,5 sin peaks and a nervous resulting line...that should not fit to the following Gray sequence.. or did I miss something?

Gunter

[Bongobat]

Hi, First sorry seems two conversations are going on. Maybe we should split this topic in the future

@ Walter

Thanks for the info on the actuators. The servos sound like a good idea too. Okay I know... I am cheap but what about this: http://www.dealextreme.com/details.dx/sku.12859 Do these work like you were describing? Also I think you are talking about pulse width modultion? If so then yes the atmega chips do have that feature

Greg

[WalterMo]

Greg,
We should really split this topic, so please do it with your next post if you see a future using a slide projector for structured light scanning.
The dealextreme servo is just the kind of servo that I mean.
I think no model boat or – aircraft can work without them. And its controlling seems to be internationally standardized. Yes, they are controlled by pulse width modulation. The atmega8 seems to be very universal.

In the past there were also suggestions on this forum to mount the laser at a servo and sweep the line by this way over the object. I think the resolution of servos is too coarse for this application.

But if such a servo is at your hand, give it a try.

Walter

[joecnc2006]

Thanks, I found a bug, now it should work. The window should cover the entire screen of the choosen Monitor. Any screen resolution and any position should work. But the user settings are still not saved and restored after restart. Please download again:

DAVID_2_4_1_CLA.zip

Simon, it works for me now, looks good.